gsd-eval-review

An AI phase evaluation review retroactively checks executed evaluation coverage, scoring dimensions as COVERED, PARTIAL, or MISSING. It diagnoses adherence to your evaluation strategy and identifies specific coverage gaps for remediation.

An EVAL-REVIEW.md is a structured document generated by reviewing evaluation artifacts in ai-evals.md. It captures dimension scores, overall verdicts, identified gaps, and actionable remediation guidance to improve future evaluation strategy adherence.

To remediate missing or partial dimensions in an AI evaluation strategy, use the generated review document to map gaps and apply the structured remediation actions. This addresses MISSING and PARTIAL scores to improve future project workflow phases.

You should run an AI evaluation review retroactively after an AI phase completes within the project workflow. This timing allows you to verify evaluation strategy adherence and inform remediation planning for subsequent phases.

Generating the EVAL-REVIEW.md with scores and remediation actions requires existing evaluation artifacts in ai-evals.md. The review process depends on these artifacts to diagnose coverage and map evaluation gaps accurately.